Академический Документы
Профессиональный Документы
Культура Документы
Nathan Rosenberg
Stanford University
October 2004
[This paper was prepared for presentation on the occasion of the opening of a
new “Center for Interdisciplinary Studies of Science and Technolo gy” at Stanford
University, November 1, 2004].
Not to prolong your suspense, the correct answer to the question in the subtitle of my
paper is the obvious one: causation runs both ways. But I want to persuade you that the
causation running from technology to science is vastly more powerful than is generally
realized.
undertake certain kinds of scientific research. This is because the eventual findings of
such research can be made to improve the performance, or to reduce the cost, of
technologies that are vital to the competitive success of profit-making firms. Further, I
want to suggest that there were powerful forces at work in the course of the 20th century
that had the effect of expanding the ways in which changes in the realm of technology
have led to changes in the various realms of science. I want to call your attention to
1
incentives , that were responsible for strengthening the causal forces that flowed from
technology to science.
In order to do this, I will need to introduce just one single bit of jargon: I will use the
term "endogenous" from the perspective of the economist and not from the perspective
of the scientist. Thus, when I refer to the endogeneity of science, I am referring to the
extent to which scientific progress has been directly influenced by the working out of the
normal forces of the market place. My justification is that I will be trying to identify forces
that emerged in the course of the twentieth century that made scientific research more
I also need to emphasize one caveat that I cannot emphasize too strongly. I am not
implicitly suggesting that the financial support of the country’s scientific research should
be left to the market place. Rather, I will be calling attention to the operation of market
forces that have become increasingly supportive of scientific research. I believe that
these developments were crucial to the rapid expansion of American industry, but that is
very different from suggesting that market forces, by themselves, were sufficient.
course of the 20th century must necessarily begin by focusing on a key organizational
2
innovation: the industrial research lab. It was these corporate labs that determined the
extent to which the activi ties of the scientific community could be made to be responsive
to the needs of the larger economy. But such a statement, by itself, cannot stand alone.
This is because these research labs depended for their effective performance, in turn,
upon a network of other institutions. These included, above all, research at universities.
Before the Second World War, university research depended heavily, for its financing,
financial support from local industry for carrying out certain classes of research, mostly
of an applied nature. This was especially true of state universities, where it was
railroads] in order to justify the imposition of taxes upon the citizens of each state. In
fact, with few exceptions, funds raised by state governments went, overwhelmingly, to
This situation was totally transformed in the post World War II period when the federal
universities became the primary locus of such research. It is important to note that the
fair to say, it has flourished, has been an organizational arrangement that has been
almost unique to the US. Unlike the situation in western Europe, where basic research
has been concentrated in government labs (Max Planck, CNRS) federal laboratories in
US have accounted for less than 10 percent of basic research (9.1% in mid 1990s).
3
A further distinctive feature of great importance in the US is the very large
commitment of private industry to scientific research that the NSF defines as basic.
Private industry accounted for slightly over 30% of all basic research in the year 2000
(probably declining slightly in last few years). Although at last count there were around
16,000 private firms that had their own corporate labs, the vast majority of these firms
conduct research of a predominantly applied nature. Only a very small number do basic
research. Nevertheless, over the years, a few of these corporate labs have conducted
research of the most fundamental nature - General Electric, IBM and, most important of
all, Bell Labs before the divestiture of AT&T in 1984. Researchers in a number of
corporate labs have won Nobel Prizes, most recently Jack Kilby, of Texas Instruments,
won the Prize for Physics, in the year 2000, for research leading to the development of
the integrated circuit [Kilby’s research received financial support from the federal
government].
Having said this , it is essential to realize that the research activities of industrial labs
should not be evaluated, as they often are by academics, by the usual academic criteria
Prizes. Such labs have a very different purpose. The industrial lab is essentially an
institutional innovation (of German origin) in which the research agenda is largely
shaped by the short-term needs but also, in a few notable cases, by the longer-term
strategies of industrial firms. Within the industrial context, the intended role of corporate
4
context of (mostly) high tech sectors of the economy. Thus, the critical achievement of
the growth of the American industrial lab in the course of the 20th century has been to
subject science, more and more, to commercial criteria. In so doing, it has rendered
science an activity whose directions were increasingly shaped by economic forc es and
One further strategic role of the corporate lab arises from the fact that a firm cannot
effectively monitor and evaluate the findings, and the possible implications, of the huge
volume of university research unless it has its own internal capability for doing such
societies that are now simply flooded with the flow of information, not only from
Internet search engines such as Yahoo and Google, the exploitation of this vast flow of
information requires an internal competence that, typically, only inhouse scientists can
provide. Indeed, America’s remarkable commercial successes in high tech markets over
the past 50 years have owed a great deal to these internal competences in private
industry. Industrial scientists have played a critical role in the transfer of potentially
useful knowledge generated by university research, not only because of their scientific
sophistication, but also because they have had a deep awareness of their firms’
5
and David Mowery and Nathan Rosenberg, Paths of Innovation: Technological
I would like now to call your attention to another major force for advancing the
endogeneity of science in the course of the 20th century. I would like to pose the
unwinding of the intertwining of science and technology suggests that the willingness of
the prospect of converting such research findings into finished and marketable products.
The actual conduct of scientific research may not be undertaken with highly specific
objectives in mind, but rather with an increased confidence that, whatever the specific
likelihood of being able to use these findings to bring improved or new products to the
market place.
From this perspective, there is a serious sense in which the economist may argue that
6
Alternatively put, the growing sophistication of engineering disciplines has had the result
sound too paradoxical. I mean to suggest that the willingness of private industry to
raises the confidence of corporate decisionmakers that the findings of basic research
field that was opened by the researches of Staudinger, Meyer and Mark in Germany in
the 1920s. In the US at least, polymer chemistry is a field that has long been dominated
deal to the increasing maturity of chemical engineering in the preceding decade or so,
Hounshell and Smith]. Carothers’ research findings led directly to the discovery of
nylon, the first of a proliferation of synthetic fibers that came to constitute an entirely
new subsector of the petrochemical industry after the Second World War. But it is
doubtful that du Pont would have committed itself to Carothers’ costly, fundamental
researches in polymer chemistry, in the first place, in the absence of the progress in
7
Nathan Rosenberg, “Technological Change in Chemicals: The Role of University-
Let me sketch out the intermediate steps that underly my argument. The discipline of
chemical engineering really had its beginnings in the second and third decades of the
20th century, mainly at MIT, in response to the spectacular expansion of the automobile
industry and, along with that industry’s growth, a voracious demand for refined chemical
products (primarily, of course, for high octane gasoline). The scale of that growth can be
captured in the following numbers: In 1900 the automobile industry was so insignificant
that the Census Bureau classified cars under the category “Miscellaneous.” [In that
year there were only 8,000 registered cars in the US]. By 1925 the automobile industry
had leaped to the status of the largest manufacturing industry in the whole country
It was the growth of the automobile that gave birth to the discipline of chemical
engineering. Chemical engineers, during the 1920s and later transformed the
petroleum refining industry from small-scale batch production into one of vastly larger
possible to introduce scientific concepts and methodologies from such fields as fluid
flow (fluid dynamics), heat transfer and, in the 1930s, the pervasive power of
thermodynamics. In other words, the design of chemical process plants could now draw
8
heavily upon a number of different scientific realms. Thus, it was the establishment of a
technology, that, in turn, laid the basis for the profitability of scientific research, not only
in du Pont and petroleum refining firms, but in a very wide range of industries that also
made use of chemical process plants. It is worth emphasizing how pervasive chemical
process plants became in the course of the 20th century. Large chemical plants could be
found in petroleum refining, rubber, leather, coal (by-product distillation plants), food-
processing, sugar refining, explosives, ceramics and glass, paper and pulp, cement,
The next related observation with respect to the growing endogeneity of scientific
research goes beyond the role played by engineering disciplines in strengthening the
private incentives to perform scientific research. The argument here is that the
development of some specific new product, that is perceived to have great commercial
potential, may provide, and often has provided, a powerful stimulus to scientific
research. This proposition is surprising only if one is already committed to a rigid, overly
simplistic linear view of the innovation process, one in which causality is always
expected to run from prior scientific research to “downstream” product design and
9
signal that a new set of profitable opportunities has been opened up in some precisely-
identified location. Consequently, it is understood that scientific research that can lead
to further improvements in that new technology may turn out to be highly profitable.
anomalous observations and unexpected difficulties that they have encountered, have
This was dramatically demonstrated in the case of the advent of the transistor, the
discovery of which was announced at Bell Labs in the summer of 1948. Within a decade
of that event solid-state physics, which had previously attracted the attention of only a
small number of researchers and was not even taught at the vast majority of American
universities (mainly MIT, Princeton, and Cal Tech) had been transformed into the
largest subdiscipline of physics. It was the development of the transistor that changed
that situation by dramatically upgrading the potential financial payoff to research in the
solid state. J.A. Morton, who headed the fundamental development group that was
formed at Bell Labs after the invention of the transistor, reported that it was extremely
difficult to hire people with a knowledge of solid-state physics in the late 1940s.
resources to perform research in the solid state occurred in the university community as
10
well as in private industry, immediately after the announcement of the momentous
findings of Shockley and his research colleagues at Bell Labs. As one strong piece of
evidence for this view, the number of publications in semiconductor physics rose from
less than 25 per annum before 1948 to over 600 per annum by the mid-1950s (Herring,
The chronology of the events that I have just referred to is essential to my argument.
Transistor technology was not the eventual consequence of a huge prior buildup of
resources devoted to solid-state physics, although it was of course also true that some
of the twentieth century’s most creative physicists had been devoting their considerable
energies to the subject. Rather, it was the initial breakthrough of the transistor, as a
functioning piece of hardware, that set into motion a vast subsequent commitment of
financial support for scientific research. Thus, the difficulties that Shockley encountered
with the operation of the early point-contact transistors led him into a systematic search
quantum physics of semiconductors. This search not only led eventually to a vastly
superior amplifying device, the junction transistor; it also contributed to a much more
and highly influential book, Electrons and Holes in Semiconductors, drew heavily upon
this research, and the book was the direct outgrowth of an in-house course that
Shockley had taught for Bell Labs’ personnel. Moreover, Shockley also found it
necessary to run a six day course at Bell Labs in June 1952 for professors from some
11
thirty universities as part of his attempt to encourage the establishment of university
Clearly, the main flow of scientific knowledge during this critical period was from
industry to university, and not the other way around. Indeed, for a considerable period
of time, Stanford and the University of California at Berkeley had to employ scientists
chemistry, after problems with the reliability of early transistors pointed in that direction.
More recently, and to compress a much more complex chain of events, the
development of laser technology suggested the feasibility of using optical fibers for
telephone transmission purposes. This possibility naturally pointed to the field of optics,
where advances in scientific knowledge could now be expected to have potentially high
great resurgence in the 1960s and after. It was converted by changed expectations,
based upon recent and prospective technological innovations, from a relatively quiet
activity in the discipline was generated, not by forces internal to the field of optics, but
fundamental research. As Harvey Brooks has noted: "While the solid-state laser gave a
new lease of life to the study of insulators and of the optical properties of solids, the gas
12
laser resuscitated the moribund subject of atomic spectroscopy and gas-discharge
physics."[Harvey Brooks, "Physics and the Polity," Science, 1968, vol. 160].
I draw the conclusion from this examination that, under modern industrial conditions,
technology has come to shape science in the most powerful of ways: by playing a major
role in determining the research agenda of science as well as the volume of resources
devoted to specific research fields. One could examine these relationships in much finer
detail by showing how, throughout the high tech sectors of the economy, shifts in the
technological needs of industry have brought with them associated shifts in emphasis in
scientific research. When, for example, the semiconductor industry moved from a
reliance upon discrete circuits (transistors) to integrated circuits, there was also a shift
chemical etching that printed the transistors on the silicon wafers and also laid down the
tracks between them. This chemical technique did away with expensive wiring, and also
produced integrated circuits that operated at much higher speeds. At the same time, the
increased reliance upon chemical methods brought with it an increased attention to the
I cite the experience of changing methods of wafer design and fabrication to indicate
the ways in which the changing needs and priorities of industry have provided the basis
for new priorities in the world of scientific research. But it is essential to emphasize that
these new priorities exercised their influence, not only upon the world of industrial
13
research, but upon the conduct of research within the university community as well. I
need only point out that Stanford University has had, for some time, its own Center for
materials, devices, and systems, and is jointly financed by the federal government and
private industry.
SERENDIPITY
would like to call your attention. I refer to the role of serendipity. It is, of course, to be
expected that well-trained scientific minds are likely to turn up unexpected findings in
concerned, chance favors only the prepared mind." By way of contrast, consider
Thomas Edison, by universal consent a brilliant inventor, but someone who had little
the flow of electricity across a gap, inside a vacuum, from a hot filament to a metal wire.
Since he saw no practical application and had no scientific training, he merely described
the phenomenon in his notebook and went on to other matters of greater potential utility
in his effort to enhance the performance of the electric light bulb. Edison was, of
course, observing a flow of electrons, and the observation has since even come to be
referred to as the "Edison Effect" - named after the man who, strangely enough, had
failed to discover it. Had he been a curious (and patient) scientist, less preoccupied
with matters of short-run utility, Edison might later have shared a Nobel Prize with Owen
14
Richardson who analyzed the behavior of electrons when heated in a vacuum, or
conceivably even with J.J. Thomson for the initial discovery of the electron itself.
Edison's "prepared mind," however, was prepared only for observations that were likely
A distinctive feature of the 20th century in dynamic capitalist economies was the
became the basis, on many occasions, for fundamental breakthroughs that occurred
the 20th century - not achieved in an industrial laboratory - was Alexander Fleming's
brilliant conjecture, in 1928, that the unexpected bactericidal effect that he had observed
in the bacterial cultures in his Petri dish, was caused by a common bread mould that
had accumulated on his slides. Fleming published this finding in 1929, but no
substantial progress was made in producing a marketable product until more than a
decade later, when the exigencies of wartime led to a joint, Anglo-American "crash"
program to accelerate the production of the antibiotic [Elder, Albert Lawrence (ed.),
It is at least a plausible speculation that, had Fleming made his marvelous discovery
while working in a pharmaceutical lab, penicillin would have become available, in large
15
quantities, far more swiftly than was in fact the case [For a contrary view, see Bernal,
volume 3, pp. 926-7]. In the context of this paper it is also worth pointing out a little-
known historical fact, that the technology to produce the antibiotic in bulk was achieved
not, as would ordinarily have been expected, by the pharmaceutical chemist, but by
chemical engineers. It was the chemical engineers who demonstrated how a technique
The growth of organized industrial labs in 20th century America vastly enlarged the
phenomena that were most unlikely to occur, or to be observed, except in some highly
specialized industrial context. In this sense, the huge increase in new high tech
Consider the realm of telephone transmissions. Back at the end of the 1920s, when
transatlantic radiotelephone service was first established, the service was discovered to
be poor due to a great deal of interfering static. Bell Labs asked a young man, Karl
Jansky, to determine the source of the noise so that it might be reduced or eliminated.
He was given a rotatable antenna to work with. Jansky published a paper in 1932 in
which he reported that he had found three sources of noise: Local thunderstorms, more
distant thunderstorms, and a third source which he described as "a steady hiss static,
16
the origin of which is not known." It was this "star noise" as Jansky labelled it, which
basic research and applied research is extremely difficult to carry out in practice.
Fundamental scientific breakthroughs often occur while dealing with very applied or
in an industrial context.
But the distinction breaks down in another way as well. It is essential to distinguish
between the personal motives of the individual researchers and the motives of the
decisionmakers in the firm that employs them. Many scientists in private industry could
honestly say that they are attempting to advance the frontiers of basic scientific
knowledge, without any concern over possible applications. At the same time, the
motivation of the research managers, who decide whether or not to finance research in
useful findings.
This certainly appears to have been the case in the early 1960s when Bell Labs
whole range of problems and possibilities in the realm of microwave transmission, and
especially in the use of communication satellites for such purposes. It had become
17
apparent that, at very high frequencies, annoying sources of interference in
This source of signal loss was a matter of continuing concern in Bell Labs'
practical concerns that Bell Labs decided to employ two astrophysicists, Arno Penzias
and Robert Wilson. Penzias and Wilson would undoubtedly have been indignant if
anyone had suggested that they were doing anything other than basic research. They
first observed the cosmic background radiation, which is now taken as confirmation of
the "Big Bang" theory of the formation of the universe, while they were attempting to
identify and measure the various sources of noise in their antenna and in the
atmosphere. It seems fair to say that this most fundamental breakthrough in cosmology
in the past century was entirely serendipitous. Although Penzias and Wilson did not
know it at the time, the character of the background radiation that they discovered was
just what had been postulated earlier by cosmologists at Princeton who had devised the
Big Bang theory. Penzias and Wilson shared a Nobel Prize in Physics for this finding.
Their findings were as basic as basic science can get, and it is in no way diminished by
observing that the firm that had employed them did so because the decisionmakers at
The parallelism between the fundamental discoveries of Jansky and Penzias and
Wilson is, of course, very striking. In both episodes, the Bell Labs researchers stumbled
18
projects that were motivated by the desire of Bell Labs to improve the quality of
telephone transmission. In the case of Penzias and Wilson, they were conducting their
research with a remarkably sensitive horn antenna that had been built for the Echo and
Telstar satellite projects. Wilson later stated that he was originally attracted to work at
Bell Labs because working in the Labs would provide access to a horn antenna which
was one of the most sensitive of such antennas in existence [Steve Aaronson, "The
Light of Creation - an Interview with Arno A. Penzias and Robert C. Wilson," Bell
discovered natural phenomena of immense scientific significance while the firm that
employed them did so in the hope that they would solve serious problems connected
with the performance of a new communications technology. In one sense it is fair to say
unintentionally - they have discovered things that they were not looking for, which I take
understand if one insists on drawing sharp distinctions between basic and applied
research on the basis of the motivations of those performing the research. I find it
irresistible here to invoke, once again, the shade of the great Pasteur: "There are no
19
In fact, I would go much further: when basic research in industry is isolated from the
become sterile and unproductive. Much of the history of basic research in American
industry suggests that it is likely to be most effective when it is highly interactive with the
work, and the concerns, of applied scientists and engineers within the firm. This is
because the high technology industries have continually thrown up problems, difficulties
and anomalous observations that were most unlikely to occur outside of specific high
technology contexts.
The sheer growth in the number of trained scientists in industrial labs, along with the
growth of new, highly complex, specialized products that appeared in the course of the
20th century, powerfully increased the likelihood of serendipitous findings. High tech
industries provide a unique vantage point for the conduct of basic research but, in order
create opportunities and incentives for interaction with other components of the firm.
Bell Labs before divestiture (1984) is probably the best example of a place where the
institutional environment was most hospitable for basic research. I do not suggest that
Bell Labs was, in any respect, a representative industrial lab. Far from it. It was a
regulated monopoly that could readily recoup its huge expenditures on research. But,
perhaps even more important, it came to occupy a location on the industrial spectrum
exploration of certain portions of the natural world that had not been previously studied.
20
INSTRUMENTATION
very modest and partial sketch. Entire categories of the influence of technology upon
science have been completely ignored here, such as the pervasive impact of new
Indeed, scientific instruments may be usefully regarded as the capital goods of the
research industry. Much of this instrumentation, in turn, has had its origins in the
university world and, to underline the extent of the intertwining of technology and
science in recent years, some of the most powerful of those instruments, such as
Nuclear Magnetic Resonance, had their origins in fundamental research that was
originally undertaken in order to acquire some highly specific pieces of knowledge, such
Bloch was awarded Stanford’s first Nobel Prize in physics for precisely such research.
the 21st Century, Edward Elgar, 1997. See also, in the same volume, Carlos
tool in chemistry for determining the structure of certain molecules (e.g., hydrogen,
21
Clearly, instrumentation and techniques have moved from one scientific discipline to
another in ways that have been highly consequential for the progress of science. In fact,
science have influenced one another. This understanding is frequently tied directly to
the development, the timing and the mode of transfer of scientific instruments among
disciplines. The flow of “exports” appears to have been particularly heavy from physics
to chemistry, as well as from both physics and chemistry to biology, to clinical medicine
and, ultimately, to the delivery of health care. There has also been a less substantial
flow from chemistry to physics and, in recent years, from applied physics and electrical
engineering to health care. NMR eventually became the basis for one of the most
physics, but the successful completion of that revolution was in turn heavily dependent
sufficiently high degree of purity and crystallinity. Finally, physics has spawned
One further point, however, is implicit in what has already been said. The availability
discipline has often been the source of interdisciplinary collaboration. In some critical
22
cases, it has involved the migration of highly trained scientists from one field to another,
such as those physicists from the Cavendish Laboratory at Cambridge University who
played a decisive role in the emergence of molecular biology. This emergence had
transferred the indispensable tool of x-ray crystallography into the very different realm of
biology. Molecular biology was the product of interdisciplinary research in the special
sense that scientists trained in one discipline crossed traditional scientific boundary
lines and brought the intellectual tools, concepts and experimental methods into the
service of an entirely new field [See the magisterial, yet highly accessible volume by
Horace Judson on the early history of molecular biology, The Eighth Day of Creation].
The German physicist, Max von Laue, discovered the phenomenon of x-ray
diffraction in 1912. Its applications were, in the early years, employed by William Bragg
and his son, Lawrence Bragg, primarily in the new field of solid-state physics but also,
later on, in developing the field of molecular biology. The main center of the
methodology of x-ray diffraction was, for many years, the Cavendish Laboratory,
presided over by Lawrence Bragg. Numerous scientists went there in order to learn how
to exploit the technique, including Max Perutz, at the time a chemist, James Watson,
Francis Crick, John Kendrew, all later to receive Nobel Prizes in Physiology and
Medicine?). The transfer of skills in x-ray diffraction was facilitated by the unusual step
Perutz but under the general direction of the physicist Lawrence Bragg [Francis Crick,
What Mad Pursuit, Penguin 1988, p. 23. James Watson later reported Bragg’s
23
obvious delight over “...the fact that the X-ray method he had developed forty
years before was at the heart of a profound insight into the nature of life itself.”
The Double Helix (Simon Schuster, 1968) p. 220].. To infer the three-dimensional
to have been a hellishly difficult enterprise, but it provided much of the basis for the new
discipline of molecular biology. Rosalind Franklin who, sadly, died very young, is widely
interacted with and influenced one another in ways that were truly symbiotic. Precisely
because these two communities marched to the tunes of very different drummers, each
was ultimately responsible for innovative improvements that could not have been
achieved by the other, had the other been acting alone [See Annetine Gelijns and
have us ually moved more readily across disciplinary boundary lines in industry than
they have in the academic world. Profit-making firms are not particularly concerned with
where those boundary lines have been drawn in the academic world; they tend to
search for solutions to problems regardless of where those solutions might be found
[NRC 1986] .
24
Thus, the technological realm has not only played a major role in setting the research
agenda for science, as I have argued. Technology has also provided new and
immensely more powerful research tools than existed in earlier centuries, as is obvious
Hubble telescope in the study of the macro-universe, and to the laser, which has
become the most powerful research instrument throughout the realm of the science of
chemistry. In addition, the laser has found a wide range of uses in medical care.
Finally, since this article has been written within easy walking distance of the Stanford
Linear Accelerator, it seems appropriate to close with the following observation: in the
realm of modern physics it appears that the rate of scientific progress has been largely
paced by technology and not by the physical laws. We always seem to ask more
25
26
27
28